Run 11. Outer Step 225. Inner Step 0.

config.json

@@ -264,8 +264,7 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "block_list": [
-    6668705,
-    6668737
+    6668770
   ],
   "bos_token_id": 1,
   "dtype": "float32",
@@ -274,7 +273,7 @@
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
-  "inner_step": 40,
+  "inner_step": 0,
   "intermediate_size": 5632,
   "last_allreduce_block": 6668039,
   "max_position_embeddings": 2048,

gradients.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ea4bffbfa186e65330dc5a6604d96305179c1170c68ae2971471c08a206e86b8
+oid sha256:1a08261cb02548d780550d250bf4136f6be7add5ebeba6436d32ac8dc23aba2b
 size 88389626

inner_optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a2128dcf5eb446e5e64bdc55e47c75df16deb3967b952dc9ee256513d33e8750
+oid sha256:bc6696a129bfc4a20176ae17e7ee320e9f16a0ea2fd8ec149f5629ff5c9be60b
 size 4924926906

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:61a317041db6e21ba1e58329472f5b77850f79e61cdef3004407685f418b9493
+oid sha256:82377fac451e4fb009bb8e97d09ff2902030101f260cd04c01700ff120925808
 size 4400216536